Dataset Information

Deep learning risk assessment models for predicting progression of radiographic medial joint space loss over a 48-MONTH follow-up period.

ABSTRACT:

Objective

To develop and evaluate deep learning (DL) risk assessment models for predicting the progression of radiographic medial joint space loss using baseline knee X-rays.

Methods

Knees from the Osteoarthritis Initiative without and with progression of radiographic joint space loss (defined as ≥ 0.7 mm decrease in medial joint space width measurement between baseline and 48-month follow-up X-rays) were randomly stratified into training (1400 knees) and hold-out testing (400 knees) datasets. A DL network was trained to predict the progression of radiographic joint space loss using the baseline knee X-rays. An artificial neural network was used to develop a traditional model for predicting progression utilizing demographic and radiographic risk factors. A combined joint training model was developed using a DL network to extract information from baseline knee X-rays as a feature vector, which was further concatenated with the risk factor data vector. Area under the curve (AUC) analysis was performed using the hold-out test dataset to evaluate model performance.

Results

The traditional model had an AUC of 0.660 (61.5% sensitivity and 64.0% specificity) for predicting progression. The DL model had an AUC of 0.799 (78.0% sensitivity and 75.5% specificity), which was significantly higher (P < 0.001) than the traditional model. The combined model had an AUC of 0.863 (80.5% sensitivity and specificity), which was significantly higher than the DL (P = 0.015) and traditional (P < 0.001) models.

Conclusion

DL models using baseline knee X-rays had higher diagnostic performance for predicting the progression of radiographic joint space loss than the traditional model using demographic and radiographic risk factors.

SUBMITTER: Guan B

PROVIDER: S-EPMC7137777 | biostudies-literature | 2020 Apr

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deep learning risk assessment models for predicting progression of radiographic medial joint space loss over a 48-MONTH follow-up period.

Guan B B Liu F F Haj-Mirzaian A A Demehri S S Samsonov A A Neogi T T Guermazi A A Kijowski R R

Osteoarthritis and cartilage 20200206 4

<h4>Objective</h4>To develop and evaluate deep learning (DL) risk assessment models for predicting the progression of radiographic medial joint space loss using baseline knee X-rays.<h4>Methods</h4>Knees from the Osteoarthritis Initiative without and with progression of radiographic joint space loss (defined as ≥ 0.7 mm decrease in medial joint space width measurement between baseline and 48-month follow-up X-rays) were randomly stratified into training (1400 knees) and hold-out testing (400 kne ...[more]

PMID: 32035934

Similar Datasets

Project description:ObjectiveTo evaluate the rate of radiographic structural progression in the sacroiliac (SI) joints in patients with radiographic or nonradiographic axial spondyloarthritis (SpA), and to determine factors predisposing to such progression, over 2 years.MethodsPatients with recent-onset axial SpA (from the Devenir des Spondyloarthropathies Indifferérenciées Récentes cohort) were assigned a radiographic SI joint score according to the modified New York criteria. Demographic characteristics, smoking status, HLA-B27 positivity, inflammation on magnetic resonance imaging (MRI) of the SI joints, disease activity, and treatment were investigated as potential predisposing factors. The main analysis consisted of the evaluation of the switch from nonradiographic to radiographic axial SpA, but other definitions of radiographic progression were also evaluated.ResultsOf the 708 patients enrolled, 449 had baseline and 2-year pelvic radiographs. Of these patients, 47% were men. Their mean?±?SD age was 34?±?9 years, 61% were B27 positive, and 37% had inflammation of the SI joints on MRI. The percentages of patients who switched from nonradiographic to radiographic axial SpA (4.9% [16 of 326]) and from radiographic to nonradiographic axial SpA (5.7% [7 of 123]) were low. The mean?±?SD change in the total SI joint score (range 0-8) was small (0.1?±?0.8) but highly significant (P?<?0.001). The potential baseline predisposing factors for meeting the modified New York criteria in the multivariate analysis were current smoking, HLA-B27 positivity, and inflammation of the SI joints on MRI, with odds ratios of 3.3 (95% confidence interval [95% CI] 1.0-11.5], 12.6 (95% CI 2.3-274), and 48.8 (95% CI 9.3-904), respectively.ConclusionOur findings suggest that structural progression does exist in early SpA, but it is quite small and observed in a small number of patients, and that environmental (smoking status), genetic (HLA-B27 positivity), and inflammation (inflammation of the SI joints on MRI) markers might be independent predisposing factors for progression.

Project description:BACKGROUND:No consensus is available regarding the best method for measuring tibiofemoral joint space width (JSW) on radiographs to quantify joint changes after injury. Studies that track articular cartilage thickness after injury frequently use patients' uninjured contralateral knees as controls, although the literature supporting this comparison is limited. PURPOSE:(1) To compare JSW measurements using 2 established measurement techniques in healthy control participants and (2) to determine whether the mean JSW of the uninjured contralateral knee in a cohort with anterior cruciate ligament (ACL) reconstruction is different from that obtained from a true control population. STUDY DESIGN:Cohort study (diagnosis); Level of evidence, 2. METHODS:Medial and lateral JSWs were measured on standardized, bilateral, semiflexed metatarsophalangeal positioning, posteroanterior radiographs of 60 healthy individuals (26 females; mean ± SD age, 25 ± 6.2 years; no history of knee injury) via 2 published techniques: a computerized surface-delineation method (surface-fit method) and a manual digitization method (midpoint method). Bland-Altman method was used to examine the agreement between JSW measurements obtained with the 2 methods and to examine the agreement between measurements obtained on left and right knees within a participant for each measurement method. Within- and between-participant variance components and intraclass correlation coefficients (ICCs) were computed for JSW measurements corresponding to each method. Two-sample t tests were used to compare the surface-fit method measurements of mean JSW of the true control group (n = 60) with the previously published mean JSW measurements from the Multicenter Orthopaedics Outcomes Network (MOON) nested cohort of 262 contralateral uninjured knees 2 to 3 years after ACL reconstruction. RESULTS:For JSW in the medial compartment, the surface-fit method had lower within-participant interknee variability (?2within, 0.064; 95% CI, 0.04-0.09) compared with the midpoint method (?2within, 0.28; 95% CI, 0.20-0.43) and a higher ICC (0.93 vs 0.65; P < .001). Lateral JSW values were similar for the surface-fit method (?2within, 0.27; 95% CI, 0.18-0.43) and the midpoint method (?2within, 0.20; 95% CI, 0.14-0.31), with ICCs of 0.75 and 0.77, respectively (P = .80). With the surface-fit method, mean JSW measurements of the medial and lateral compartments of a control population were not significantly different from the contralateral uninjured knees of patients after ACL reconstruction. CONCLUSION:For measuring medial JSW, the surface-fit method was less variable across knees within a participant than the midpoint method, as evidenced by larger ICCs and lower interknee variability. For measuring lateral JSW, the 2 methods were similar. The JSW measurements of uninjured contralateral knees of patients with ACL reconstruction at 2 to 3 years postsurgery were not significantly different from those of a cohort of healthy control participants. Future work should be performed to demonstrate the validity of these methods for documenting change over time in the ACL-reconstructed knee.

Project description:Minimally invasive surgery (MIS) represents one of the most innovative surgical treatments of hallux valgus (HV). However, long-term outcomes still remain a matter of discussion within the orthopaedic community. The purpose of this longitudinal prospective study was to evaluate radiographic and functional outcomes in patients with mild-to-severe HV who underwent Reverdin-Isham and Akin percutaneous osteotomy, following exostosectomy and lateral release.Eighty patients with mild-to-severe symptomatic HV were treated by MIS. Clinical evaluation was assessed preoperatively, as well as at 3 and 12 months after surgery and at final follow-up of 48 months, using the American Orthopaedic Foot and Ankle Society (AOFAS) hallux grading system. Patient satisfaction and complications were recorded. Computer-assisted measurement of antero-posterior radiographs was taken preoperatively, as well as at 3 and 12 months after surgery and at 48-month follow-up, analysing the intermetatarsal angle (IMA), the hallux valgus angle (HVA), the distal metatarsal articular angle (DMAA) and the tibial sesamoid position. Also, the bridging bone/callus formation was evaluated at the different radiographic follow-ups, while the articular surface congruency and the metatarsal index were calculated only preoperatively and at the last follow-up. Patient satisfaction was assessed using the visual analogue score (VAS). Statistical analysis was carried out using the paired t test. Statistical significance was set at p?<?0.05.The mean AOFAS score was 87.15 points at the final follow-up of 48 months, and the VAS score was 8.35/10. The post-operative radiographic assessments showed a statistically significant improvement compared with preoperative values. The mean corrections of each angular value at the last follow-up were as follows: IMA 3.90°, HVA 12.50°, DMAA 4.72° and a tibial sesamoid position of 1.10. The articular surface was congruent in 77 (96.25%) cases and incongruent only in 3 (3.75%). The complete healing of the osteotomies was achieved in all series at 3-month follow-up. However, the results obtained in the correction of the severe HV deformities were less encouraging.Minimally invasive surgery with Reverdin-Isham and Akin percutaneous osteotomy, in combination with previous exostosectomy and subsequent lateral soft-tissue release, is a safe, effective and reliable procedure for correction of mild-to-moderate HV. However, it requires a long learning curve because of the inherent difficulty of the mixed different surgical procedures.ClinicalTrials.gov PRS Protocol Registration and Results System: NCT02886221.

Dataset Information

Deep learning risk assessment models for predicting progression of radiographic medial joint space loss over a 48-MONTH follow-up period.

Objective

Methods

Results

Conclusion

Publications

Deep learning risk assessment models for predicting progression of radiographic medial joint space loss over a 48-MONTH follow-up period.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets